Optical fibers have been used mainly for optical communication systems for decades. Recently, optical fiber sensing technologies have grown rapidly due to their advantages over conventional sensing devices. The optical fiber sensors can handle much higher signal bandwidth and temperature ranges, are immune to electromagnetic interference noises, provide safe operation (no generation of electric sparks which occasionally induce catastrophic fire incidents), and require a much easier installation process. But the most prominent feature is their capability of true distributed parameter measurement i.e., the extended ranges' temperature monitoring can be covered with a single fiber optic cable. Utilizing this distributed technology, temperature or other parameter's profile along significant distances can be monitored continuously without any electric wire involvement. Many temperature data points can be processed along over 10 kilometers within a short time. The resultant distributed measurement is equivalent to the performances of thousands of ‘point temperature’ sensors, which usually are susceptible to electric noises, occasional fire incidents due to wire shorted circuits and complicated installation—costly and take long periods of time. Thus the fiber optic distributed sensing system provides for long range temperature measurement with reliable performance, safe operation and economic installation.
When a laser light with a center wavelength (1) is injected into a fiber cable, most of the light is transmitted, but small portions of incident lights are scattered backward or forward along the fiber. These scattered lights are categorized into three unique spectral bands—Rayleigh, Brillouin and Raman bands. For the measurement of distributed temperature, typically few components such as Rayleigh (1R, insensitive to temperature), Stokes (1S, longer than 1R and less sensitive to temperature) and anti-Stokes (1AS, shorter than 1R and most sensitive to temperature) of Raman band have been used. These optical signals may be separated by optical filters (or other wavelength selecting devices) and are received by the photo detectors to convert light to electrical signals. The temperature can be calculated by the ratio of temperature sensitive anti-Stokes to less sensitive Stokes or temperature insensitive Rayleigh components.
To obtain the temperature profile along the distance, two processing methods—time domain processing approach and frequency domain approach have been applied conventionally. The time domain method (or OTDR—Optical Time Domain Reflectometry) uses pulsed light source, and the location of the temperature is identified by the calculation of the pulse's round trip time to the distance under test. The frequency method (OFDR—Optical Frequency Domain Reflectometry) uses a modulated laser source, and the position can be calculated by applying the inverse Fourier transformation of the sensing fiber's transfer function or the frequency response. The OTDR method is a one step process and provides quicker response time but needs high pulse energy to obtain high SNR (signal to noise ratio). OFDR method takes longer process time because it is a two step process (convert from frequency domain to time domain), but higher light power can be applied to increase the SNR. Also random pulse coded (multiple pulses) based on time domain methods can be applied. All approaches have their pros and cons, and the selection may depend on the application.
Even though the DTS has been widely applied to many areas so far, two critical issues should be handled properly for reliable long term measurement. The first issue is the inherent characteristics of anti-Stokes band, which are not only sensitive to temperature variation but also to physical perturbations, which induce the attenuation of the light transmitting along the fiber. This ambiguity can be corrected by using another reference light source(s), whose Rayleigh band 12R (only sensitive to the attenuation only) is located in Stokes or anti-Stokes band of the first light source, 11AS. The second issue is the DAF (Differential Attenuation Factor) due to the material characteristics of optical fiber. For given external perturbations, all transmitting light in the fiber cable including wavelengths of 1R, 1S and 1AS experience different attenuations. Typically, the shorter the operating 1 is, the higher the attenuations are. These phenomena should be considered and handled properly for reliable temperature calculation by the ratio of 1AS to 1S or 1AS to 1R bands. Typically this effect is corrected just after the sensing fiber is deployed. But in many real application cases, various perturbations are applied or generated unintentionally to the section of sensing fiber cable times after deployment. These environmental effects are called the ‘darkening’ in a local section or sections of the fiber and induce different attenuation values to each 1, resulting in erroneous temperature profile. Those effects may be: radiation related darkening, hydrogen gas related darkening and additional stresses applied on the arbitrary length of the fiber cable. Therefore this phenomenon should be corrected automatically (or self corrected) and continuously for accurate measurement. To correct DAF related errors, the DE (Dual End) method has been applied, which is implemented by two channel configuration (Ch1 and Ch2 are used for a sensing) and utilizes common loss compensation method between these 2 channels. After the primary laser light is injected to Ch1 and Ch2 consecutively by a switching device, the detected signal of Ch2 (or either one) is rotated in mirror image and subtracted (common attenuations) from the other channel, removing the error. The issue of this method is the requirement of two channels, which requires twice the measurement time, double the length of sensing fiber and twice the optical budget. Also DE configuration is not permissible for some critical applications. Finally, the temperature effect of the Stokes band is subtracted for high temperature measurement because the Stokes band is still dependent on temperature even though it is not as sensitive as the AS band.
The major trend of temperature calculation has been utilizing the ratio (R) of AS to S band because their light intensities (IS and I AS) are comparable to each other. It is expressed mathematically as:
      R    ⁡          (      T      )        =                    I        AS                    I        S              =                            (                                    λ              S                                      λ              AS                                )                4            ⁢              exp        ⁡                  (                      -                          hcv                                                k                  B                                ⁢                T                                              )                    where h, c, kB, v and T are Plank constant, the speed of light, Boltzmann constant, Raman wave number shift and absolute temperature in degree K, respectively. A similar equation can be applied for other ratios such as AS to R, both are in same or different bands. The correction method using AS/R ratio instead of AS/S ratio is more effective in terms of the amplitude of DAF (approximately half of AS/S case by an interpolation method) and the smallest temperature dependency of R band. But the amplitude of R is usually a few orders larger than AS band, and it is necessary to correct DAF continuously during all the measurement periods. The challenging job is the process of continuous in-situ correction for DAF and the subtraction of the temperature dependency of S band in high temperature application.
The disclosure in U.S. Pat. No. 7,126,680 described the correction of attenuation related effect of 11S and 11AS by two independent extra reference light sources 12 and 13 (total 3 light sources are required). In this case, several conditions need to be considered to be an effective correction—1) two extra light sources center 1 and their bandwidths should be identical and kept stable. 2) Scattered signals such as anti-Stokes or Stokes bands are much wider than extra laser sources' Rayleigh bands. Also the intensities of two reference light sources should be comparable and compensated to suppress the error related to the momentary fluctuations.
Finally, the temperature effect of the Stokes band is subtracted. The other correction algorithm using two light sources was disclosed in U.S. Pat. No. 4,767,219. In this disclosure, two light sources need to be selected to satisfy the condition that 1/11=1/12=1/v where v is Raman shift in wave number. Another scheme was disclosed in U.S. Pat. No. 7,628,531 by selecting two light sources such as 11AS=12S. For these two light source cases, DAF issue can be corrected automatically because two bands are located in the same center wavelength. But their bandwidth size as well as center wavelengths should be precisely matched and kept stable to ensure the effective correction continues. Also the intensities of secondary light sources should be stabilized to reduce temperature error related to the momentary fluctuations of the source. Last, the nature of temperature dependency of the S band should be subtracted. Because of various issues mentioned so far, the ideal correction method is to use the ratio between AS and R, which has the same wavelength bands with comparable amplitude. For continuous correction approaches, the following one source, two sources and 3 sources method have been disclosed as described below.
The idea using AS band to R band ratio with three sources was disclosed in U.S. Pat. No. 7,284,904. In this scheme, 3 lights 11 (primary light), 1-1 and 11LO (two auxiliary lights) were proposed for the scheme. 11LO is the light source with same wavelength as 11 but has lower optical power and 1-1 source is same wavelength as 11AS generated by 11. DAF correction was claimed by two auxiliary spectral bands 11LO and 1-1 by normalization and interpolation. But interpolation is an indirect correction method, and there may be a practical implementation limit to match and keep the wavelengths stable. Another auto correction method between the anti-Stoke band and R band with one light source was disclosed in U.S. Pat. No. 7,350,972. To generate R band match to AS band, a semiconductor laser was operated in two modes, from a laser mode (stimulated emission) to an LED mode (spontaneous emission) by applying the driving current under threshold level consecutively. In the LED mode, the spectral width is widened but the light intensity is also significantly decreased under impractical level. This simple auto-correction scheme has two concerns: 1) LED mode's spectral band should be wide enough to cover both whole AS band and its original R band, which is separated around 50 to 100 nm in a single side of the band (depends on operating 1, 800 nm to 1550 nm) from R band, and 2) its output power of LED mode should be high enough to be a practical implementation.